UNIVERSIDAD SAN FRANCISCO DE QUITO

Colegio de Ciencias e Ingenierías

Verification of Flexural Stop Criteria for Proof Load Tests on Concrete

Bridges Based on Beam Experiments

Proyecto de investigación

Andrés Rodríguez Burneo

Ingeniería Civil

Trabajo de titulación presentado como requisito

para la obtención del título de

Ingeniero Civil

Quito, 8 de mayo de 2017

2

UNIVERSIDAD SAN FRANCISCO DE QUITO USFQ

COLEGIO DE CIENCIAS E INGENIERÍAS

HOJA DE CALIFICACIÓN

DE TRABAJO DE TITULACIÓN

Verification of Flexural Stop Criteria for Proof Load Tests on Concrete

Bridges Based on Beam Experiments

Andrés Rodríguez Burneo

Calificación:

Nombre del profesor, Título académico Fabricio Yépez, Ph.D.

Firma del profesor ________________________________

Quito, 8 de mayo de 2017

3

© Derechos de Autor

Por medio del presente documento certifico que he leído todas las Políticas y Manuales

de la Universidad San Francisco de Quito USFQ, incluyendo la Política de Propiedad

Intelectual USFQ, y estoy de acuerdo con su contenido, por lo que los derechos de propiedad

intelectual del presente trabajo quedan sujetos a lo dispuesto en esas Políticas.

Así mismo, autorizo a la USFQ para que realice la digitalización y publicación de este

trabajo en el repositorio virtual, de conformidad a lo dispuesto en el Art. 144 de la Ley

Orgánica de Educación Superior.

Firma del Estudiante: _____________________________________

Nombres y Apellidos: Andrés Rodríguez Burneo

Código de estudiante: 00113944

Lugar, Fecha: Quito, mayo 2017

4

RESUMEN

Los ensayos de prueba de carga permiten a los ingenieros determinar si una estructura

aún es segura para su uso. Sin embargo al someter la estructura a este tipo de ensayos, esta

puede sufrir daños irreparables. Para evitar esta situación, códigos de construcción y guías

para ensayos de carga han establecido criterios para detener el ensayo antes de generar daño

irreversible. Estos criterios a menudo se basan en datos recolectados a medida que el ensayo

de carga se está llevando a cabo No obstante, estos criterios deben ser revisados a fin de

mejorarlos de manera que los ensayos puedan ser llevados a cabo de manera segura. Además

de los criterios existentes en los respectivos códigos de construcción se han presentado nuevas

propuestas de criterios de parada a fin de mejor la seguridad con que se realiza los ensayos de

prueba de carga. Este reporte analiza los resultados obtenidos de 4 experimentos en 2 vigas

fabricadas y ensayadas en laboratorio y compara estos resultados con los valores límite

obtenidos en base a los criterios de parada establecidos en el código ACI 437.2M-13 y la guía

alemana establecida por DAfStB. Adicionalmente un nuevo criterio de parada, propuesto por

Werner Vos de TU Delft en Holanda, también es comparado con los resultados

experimentales. Esta investigación busca analizar en qué circunstancias es mejor aplicar un

criterio de parada respectivo, que deficiencias tienen y que tan confiable y conservador es cada

criterio para ser aplicado no solo en edificios sino también en puentes de hormigón armado. Se

encontró que respecto a los criterios establecidos por ACI, el protocolo de carga del ensayo es

imperativo para obtener resultados confiables adecuados para ensayos de prueba de carga. En

cuanto a las otras propuestas, dependiendo del nivel de seguridad que se busca, se encontraron

resultados consistentes y confiables. Se espera que con más investigación respecto a criterios

de parad basados en flexión se puedan desarrollar mejores formas de determinar los valores

máximos admisibles lo que permitirá aplicar de una forma más segura los ensayos de pruebas

de carga.

Palabras clave: flexión, criterio de parada, ensayos de prueba de cargas, deflexión,

ancho de agrita, deformación del hormigón, rigidez del hormigón dañado.

5

ABSTRACT

Proof load tests allow engineers to determine if a structure is still suitable for use.

However, as the structure is subjected to this test it may suffer irreparable damage. To avoid

this scenario, building codes have established stop criteria for proof load tests. These stop

criteria often refer to data that is taken as the test is being carried out. However stop criteria

need to be revised in order to be improved. Additionally, other proposals of stop criteria have

been submitted to improve safety of proof load tests. This report analyses the results obtained

from 4 experiments on 2 cast-in-laboratory beams, and compares them to the values obtained

with the stop criteria established by the ACI 437.2M-13 and the German guidelines of the

DAfStB. Additionally, a new proposal for stop criteria by Werner Vos from TU Delft in the

Netherlands is also compared to the experimental results. This research aims to analyze under

which circumstances it is better to apply a specific stop criterion, which are the flaws on the

criteria from the codes and the new proposal, and how reliable they are to be applied not only

on buildings but on concrete bridges. It was found for the ACI stop criteria, that the loading

protocol is imperative to have consistent results and perform adequate proof load tests. As for

the other proposals, depending on the margin of safety considered to avoid permanent damage,

reliable results were found. Hopefully, further investigation in flexural stop criteria would help

to develop better ways to calculate the maximum allowable values, which will lead to a better

and safer application of proof load tests.

Key words: Flexure, Stop Criteria, Proof load test, deflection, crack width, concrete

deformation, damaged concrete stiffness.

6

TABLE OF CONTENTS

Resumen……………………………………………………………………………………….4

Abstract………………………………………………………………………………………..5

Introduction…………………………………………………………………………………...9

Literature Review……………………………………………………………………………12

Investigation design and methodology……………………………………………………..24

Analysis of results……………………………………………………………………………42

Conclusions…………………………………………………………………………………..66

References……………………………………………………………………………………69

Appendix……………………………………………………………………………………..70

7

TABLE INDEX

Table 1, Values of the experiment layout…………………………………………………..26

Table 2, Comparison on calculated and real ultimate load……………………………….41

Table 3, Concrete strain measured on P804A1 at every load step………………………..42

Table 4, P804A1 crack width. ………………………………………………………………43

Table 5, P804A1 linearity index…………………………………………………………….44

Table 6, P804A1 permanency ratio…………………………………………………………44

Table 7, P804A1 deflection measured at every step………………………………………..45

Table 8, Concrete strain measured on P804A2 at every load step………………………..46

Table 9, P804A2 crack width. ………………………………………………………………46

Table 10, P804A2 linearity index……………………………………………………………47

Table 11, P804A2 permanency ratio………………………………………………………..48

Table 12, Deflection P804A2 measured at every step……………………………………...48

Table 13, Concrete strain measured on P804B at every load step………………………..49

Table 14, Deflection P804B measured at every step……………………………………….50

Table 15, Concrete strain measured on P502A2 at every load step………………………50

Table 16, P502A2 crack width. ……………………………………………………………..51

Table 17, P502A2 linearity index……………………………………………………………52

Table 18, P502A2 permanency ratio………………………………………………………..52

Table 19, Deflection measured on P502A2 at every step………………………………….53

Table 20, Comparison of deflection values from Vos’s proposal…………………………53

Table 21, Comparison of crack width values from Vos’s proposal………………………54

Table 22, comparison of concrete strain on every experiment …………………………...55

Table 23, Comparison of maximum crack width on every experiment…………………..58

Table 24, Comparison of minimum crack width on every experiment…………………..58

8

FIGURES INDEX

Figure 1, ACI cyclic load stop criteria: (a) Deviation from linearity index; (b)

Permanency ratio……………………………………………………………………………18

Figure 2, ACI cyclic loading protocol………………………………………………………18

Figure 3, Moment curvature diagram, semi-linear approach…………………………….20

Figure 4, Beam experiments layout…………………………………………………………25

Figure 5, Loading scheme: (a) P804A1; (b) P804A2; (c) P804B (d) P502A2……………..27

Figure 6, P804A1 experiment results: (a) load-displacement; (b) Load-strain; (c) load-

crack width; (d) time-crack width…………………………………………………………..29

Figure 7, P804A2 experiment results: (a) load-displacement; (b) Load-strain; (c) time-

strain (d) load-crack width; (e) time-crack width………………………………………….31

Figure 8, P804B experiment results: (a) load-displacement; (b) Load-strain……………32

Figure 9, P502A2 experiment results: (a) load-displacement; (b) Load-strain; (c) time-

strain (d) load-crack width; (e) time-crack width………………………………………….34

Figure 10, Experiment layout for Vos’s proposal calculations……………………………35

Figure 11, Stiffness from retrograde branch……………………………………………….36

Figure 12, Average crack spacing according to Van Leeuwen ……………………...……65

9

INTRODUCTION

Existing civil structures deteriorate with time due to the continuous loading and

environmental conditions they are subjected to, or it can be that a structure has suffered severe

damage due to any accident. This causes a loss of their initial properties and consequently

large uncertainties on the structural behavior. Therefore analyses should be carried out in order

to confirm that the structure is still safe for use. If there is background data about the structure

to be tested, simulations and computer analyses can be done. However, the level of assessment

of these analyses may not be as close to reality as needed since the level of damage and

deterioration is not a hundred percent clear. An option for analyzing deteriorated or damaged

structure with or without background data is load testing in which the actual structure is

loaded and its behavior is measured. There are two types of load testing: the first one is

diagnostic load testing, in which the structure is loaded in order to obtain its mechanical

properties or to update analytical models. The second type is proof load testing, which is the

subject of this research. Its purpose is to assure the safety of a structure by subjecting it to a

specific maximum load known as target load. If the structure withstands the target proof load,

it passes the test and is still suitable for use.

As a structure is heavily loaded during a proof load test, it can easily get damaged before

reaching the target load since its approximate resistance is unknown due to the lack of

background data and, the level of deterioration and damage. Therefore, parameters with

threshold values must be defined in order to identify that further loading would induce

permanent damage to the structure and the test must be stopped immediately. These

parameters are the so-called “stop criteria”.

10

Since the stop criteria must be evaluated as the test is being carried out, parameters related to

ductile failure, such as deformation, deflections or changes in stiffness, can easily be

measured, thus, they are convenient when establishing stop criteria. On the other hand, brittle

failure caused by shear which is instantaneous can be used for stop criteria as well but it is out

of scope of this report.

Some building codes such as the German guidelines (Deutcher Ausschuss für Stahlbeton,

2007) and the American ACI 437.2M-13 code (ACI Committee 437, 2013), already establish

stop criteria. Nonetheless, these criteria were developed to be applied to buildings, not bridges.

Therefore, several studies and experiments have been carried out in order to improve existing

stop criteria, such as the investigation made by Werner Vos (Vos, W., 2016) at TU Delft,

which is the one analyzed in this report. The stop criterion proposed by Vos aims to establish

theoretical threshold values prior to performing the test. The procedure to obtain this value is

derived from Monnier’s (Monnier. Th., 1970) investigation on the relation between bending

moment and curvature.

Bridges represent previously loaded structures, already cracked and with a residual existing

deformation. Since they are civil structures used by hundreds of people, old bridges must be

tested with a level of assessment that can assure they are still suitable for use, or they must be

repaired or replaced immediately. This level of assessment can be reached through proof load

testing, always protecting the structure from permanent damage with its respective stop

11

criterion. Since existing stop criteria has been developed for buildings, not bridges, it needs to

be studied and reevaluated so that proof loading tests can be carried out in a safer way.

This investigation looks at revising the stop criteria mentioned above, and compares it with the

stop criterion proposed by Vos. Results obtained from beams cast and tested in the laboratory

are used to analyze the level of safety and accuracy of the existing stop criteria from the

German guideline and ACI 437.2-M13 and Vos proposal as well.

12

LITERATURE REVIEW

In order to understand the parameters used to compare experimental results in this

investigation, a brief explanation of what is proof loading, and what are the existing stop

criteria is presented.

General Aspects of Proof Load Testing

A proof load test is a test carried out on both new and old structural elements in order to assure

their safety. The objective of this test is to load a structure gradually until it reaches a

maximum specific load, known as the target load, which proves that the structure is suitable

for use.

Proof loading is a common practice when there is not enough background information to

perform a structural analysis, after the structure has been subjected to loads it was not

designed to withstand, or when it has suffered severe damage or material degradation.

Therefore, checking if the structure is able to bear a specific load allows to determine if the

structure must be repaired or replaced.

Three parameters must be established before the test is carried out: the loading protocol, the

target load and the stop criteria.

The loading protocol establishes how the test is going to be performed. This includes how and

where the loads will be applied. Loads can be in cycles, increasing the maximum applied load,

13

or in a monotonic way, increasing the load continuously after established periods of time. A

single test may include different load cases with their respective parameters. The position of

the load aims to recreate the most unfavorable condition.

The stop criteria are parameters established to protect the integrity of the structure during the

proof load test. As the structure is tested in order to reach the target load, stresses on the

structure may increase to the point at which the structure suffers permanent damage, or in the

worst case scenario, it collapses. To avoid this, stop criteria must be established, usually as

parameters that will be measured on the structure while the test is being performed. Existing

stop criteria, which will be discussed later in this report include parameters such as: maximum

deflection, crack width, deviation of linearity index among others. If one of these parameters

is exceeded during the test, it must be aborted immediately, whether the target load has been

reached or not.

There are many ways for the load to be applied. Tanks continuously filled with water can be

used for monotonic load protocol, of the BELFA truck from Germany designed to apply loads

to the deck that can be easily controlled and monitored. However, BELFA truck has a

maximum load that cannot be increased. (Koekkoek, R., 2015). A common practice in the

Netherlands is a system of hydraulic jacks in which all the weight available is placed over a

surface that transmits the load directly the supports before loading is started. During the tests

the jacks gradually transfer the load to the superstructure as they push down the surface.

(Lantsoght, E., 2016)

14

Sensors must be installed in order to obtain as much information as possible from the

structure. First of all, the loading process must be controlled. LVDTs and lasers are placed at

strategic points in order to measure displacements and deformations. Sometimes, acoustic

emission signals are measured as well, in order to relate their results to cracking. These

measurements allow the engineer who follows the measurements to identify if any stop

criterion has been exceeded.

Existing Stop Criteria

German Guideline for Proof Load Testing DAfStB Richtlinie

The German guideline for proof load testing was established in 2000, and applies to both plain

concrete and reinforced concrete structures. The protocols established rely on a ductile failure

mode. Load testing of shear-critical structures or elements is not permitted.

For proof load tests, cyclic loading must be carried out with at least 3 steps of loading and

unloading. The maximum load at which a stop criterion has been reached is defined as 𝐹𝑙𝑖𝑚.

(Deutcher Ausschuss für Stahlbeton, 2007)

Parameters established for stop criteria are:

o Concrete strain

𝜀𝑐 < 𝜀𝑐 𝑙𝑖𝑚 − 𝜀𝑐0 (1)

Where 𝜀𝑐 is the measured strain during the proof load test, 𝜀𝑐 𝑙𝑖𝑚 is the limit value for

concrete strain based on concrete characteristic compressive strength defined by the

German guideline as 0.006%, which can be increased to 0.008% in concrete

15

compressive strength is greater than 25MPa, and 𝜀𝑐0 is the short term strain caused in

the concrete by the permanent loads, determined analytically.

o Strain in reinforcement steel

𝜀𝑠2 < 0.7𝑓𝑦𝑚

𝐸𝑠− 𝜀𝑠02

𝜀𝑠2 < 0.9𝑓0.01𝑚𝐸𝑠

− 𝜀𝑠02

(2)

(3)

𝑓𝑦𝑚is the average yield strength of steel, 𝑓0.01is the average yield strength based on a

strain of 0.01%, 𝐸𝑠 is the modulus of elasticity of the steel, 𝜀𝑠2 is the steel strain

during experiment, and 𝜀𝑠02 is the analytically determined strain caused by the

permanent loads

The second equation (3) may be used when the stress-strain relationship of the steel is

known completely.

o Crack width and increase in crack width

The crack width of new cracks formed during and after the test is limited to:

𝑤 ≤ 0.5𝑚𝑚 during proof loading

≤ 0.3𝑤 after proof loading.

For previously formed cracks, their increase in width is limited as follows:

∆𝑤 ≤ 0.3𝑚𝑚 during proof loading

≤ 0.2∆𝑤 after proof loading.

o Deflection

16

Test must be stopped if more than 10% permanent deformation occurs after removing the

load, or if there is a clear increase of the nonlinear part of the deformation.

o Deformation in the shear span of beams with shear reinforcement

The test must be stopped if 60% of the concrete strain 𝜀𝑐 is reached at the concrete

compressive struts.

The test must be stopped if 50% of 𝜀𝑠2 occurs in the shear reinforcement.

American Code ACI 437.2M-13

As stated in the code: ‘’The purpose of this code is to establish the minimum requirements for

the test load magnitudes, load test procedures, and acceptance criteria applied to existing

concrete structures as part of an evaluation of safety and serviceability to determine whether

an existing structure requires repair and rehabilitation’’ (ACI Committee 437, 2013). Only the

acceptance criteria will be discussed. For further information, refer to the code itself.

Chapter 6 of the code does not establish how to determine a value for stop criteria, but

explicitly defines quantitative rules to determine if the structure passes the load test, known as

acceptance criteria. Acceptance criteria describe the acceptable limits of performance

indicators, and thus serve the same purpose as stop criteria. The codes defines qualitative

requirements as well related to the observation of cracks that could indicate failure, but it is

out of the scope of this research.

Parameters for acceptance criteria established in the code are:

17

Monotonic load protocol, deflection limits:

As established on the equations 6.3.1 and 6.3.2 respectively

∆𝑟 ≤∆𝑙4

∆𝑙 ≤𝐿

180

(4)

(5)

∆𝑟 is the residual deflection measured 24 hours after the removal of the load and ∆𝑙

represents the maximum deflection, and L is the span length

Cyclic load protocol, deviation of linearity index 𝐼𝐷𝐿:

Based on a hysteretic model, the deviation of linearity index analyses the variation of

the slope in a load-deflection plot for every loading cycle, which is calculated as

established on equation 6.4.1

𝐼𝐷𝐿 = 1 −tan(𝛼𝑖)

tan(𝛼𝑟𝑒𝑓)< 0.25

(6)

𝛼𝑖 is the secant stiffness of a point i in the loading section of the plot and 𝛼𝑟𝑒𝑓 is the

slope of the secant of the load-deflection envelope, as shown in Figure 1(a)

Cyclic load protocol, permanency ratio 𝐼𝑝𝑟:

According to the set of equations 6.4.2

𝐼𝑝𝑟 =𝐼𝑝(𝑖+1)

𝐼𝑝𝑖< 0.5

𝐼𝑝𝑖 =∆𝑟𝑖

∆𝑚𝑎𝑥𝑖

(7)

(7.1)

18

𝐼𝑝(𝑖+1) =∆𝑟𝑖+1

∆𝑚𝑎𝑥𝑖+1

(7.2)

Where i represents the number of the cycle, ∆𝑟𝑖 is the deflection shown at minimum

load 𝑃𝑚𝑖𝑛 and ∆𝑚𝑎𝑥𝑖 for 𝑃𝑚𝑎𝑥 at the i th cycle of loading. The permanency ratio is

acceptable if it does not exceed 0.5 for every pair of cycles. Data for determining Ipr is

taken as seen on figure 1(b). ACI 437.2M-13 also defines a cyclic loading protocol

shown in Figure 2.

Figure 1. ACI Cyclic loading protocol stop criteria: (a) Deviation from

linearity index; (b) Permanency ratio. (ACI Committee 437, 2013)

Figure 2. ACI Cyclic loading protocol. (ACI Committee 437, 2013)

19

Werner Vos proposal:

As a TU Delft investigation, Werner Vos proposed two stop criteria. The first one is based on

the relation stiffness-deflection, developed through a theoretical approach based on the

theoretical moment-curvature relation developed by Monnier (Monnier. Th., 1970). The

second proposal is based on the relation between crack width and deformation as developed by

Van Leeuwen. (Van Leeuwen. J., 1962)

o Stiffness-deflection proposal:

This proposal starts from the moment-curvature diagram established by Monnier, which

plots the bending moment at a section of the beam against its curvature. The slopes of the

lines in the plot represent the stiffness of the element at different stages of the concrete:

un-cracked and cracked. These values of the stiffness can be calculated using the element

dimensions and the percentage of steel reinforcement. It must be mentioned that the plot is

simplified to be semi linear, with straight lines with different slopes, in order to have

constant stiffness in-between stages. In reality, the plot is curved since the stiffness

continuously changes as concrete keeps cracking and the steel yields.

Monnier established the moment-curvature diagram for both first time and alternate

loading. The alternate loading model resembles the cyclic loads applied during a proof

load test and; additionally it allows to find the maximum applied load in the history of the

specimen and its residual deformation. Therefore, this model suits the circumstances under

which a bridge is subjected to a proof load test: with existing cracks and with an existing

residual deformation.

20

As concrete continues to crack due to bending, its stiffness decreases with every cycle in

relation to the curvature of the beam. Therefore it is a convenient parameter to include in

stop criteria. However, measuring curvature during a test requires more complex

equipment compared to the usually measured parameters: deformation, deflection and

cracks. Therefore a relation between load, stiffness, moment curvature and deflection is

established:

𝑘 =𝑑2𝛿(𝑥)

𝑑𝑥2

(8)

where k is the curvature, and δ(x) the deflection.

Vos defines a semi-linear model, as shown in Figure 3 in which the concrete element has

two stiffness: the un-cracked stiffness, EIo, and cracked stiffness from the retrograde

branch EIte, and a two-step calculation is done; one before the cracking moment is

reached, and one afterwards, using its respective stiffness. With a semi-linear approach

the relation for elastic materials can be applied:

Figure 3. Moment curvature diagram,

semi linear approach. (Vos, 2015)

21

Once the relations are established the stop criteria go as follows:

If a moment at a certain load level overpass the maximum deflection it means the beam is

yielding and the process must be stopped.

If there is a residual deformation after a cycle, larger than existing deformations caused by

self-weight and previous loads, once the cracking moment has been reached, it means

yielding has occurred and the test must be stopped. However residual deformation for

preloaded existing structures should be considered when checking for residual

deformations during the test. This values if possible can be determined with the load

history of the structure or can be measured prior to start the test

Additionally, building codes establish maximum allowable deflections based on the

element span. If this deflection is reached the test must be stopped. However, these values

are meant to be used not on bridges but buildings.

o The second proposal is based on the relation between crack width and deformation as

identified by Van Leeuwen:

𝑤𝑚𝑎𝑥 = 𝛽 6.12 𝑓𝑦 𝑠 10−6 𝑚𝑚

𝑤𝑟𝑒𝑠 = 𝛽 6.12 𝜎𝑠1 𝑠 10−6 𝑚𝑚

(9)

(10)

β is the ratio between the permanent or cyclic load and total load. In the worst case

scenario this ratio will be one. Therefore according to the relation in the formula the

maximum crack width can be found when β=1. 𝜎𝑠1 is the steel stress right at the crack,

and s is the space between cracks whose equation must be in accordance to the type of

bar: ribbed or plain.

22

Additionally, Vos includes minimum residual and maximum allowable crack width

from the Eurocode, however this does not establish any difference between plain and

ribbed bars. These values will be calculated in this report to have more results for

comparison

𝑤𝑚𝑎𝑥,𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒 =1

2

𝑓𝑐𝑡𝑚𝜏𝑏𝑚

∅

𝜌𝑒𝑓𝑓

𝑓𝑦 − 𝑘𝑡𝑓𝑐𝑡𝑚𝜌𝑒𝑓𝑓

(1 − 𝛼𝑒𝜌𝑒𝑓𝑓)

𝐸𝑠

𝑤𝑟𝑒𝑠,𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒 =1

2

𝑓𝑐𝑡𝑚𝜏𝑏𝑚

∅

𝜌𝑒𝑓𝑓

𝜎𝑠 − 𝑘𝑡𝑓𝑐𝑡∗

𝜌𝑒𝑓𝑓(1 − 𝛼𝑒𝜌𝑒𝑓𝑓)

𝐸𝑠

(11)

(12)

Where fctm is the concrete mean tensile strength, τbm is bond between reinforcement and

concrete, Ø is the bar diameter, ρeff is effective reinforcement ratio calculated only with

area of concrete under tension, kt equals 0.4 or 0.6 depending if loads applied are long

term or short term respectively, αe is the ratio between steel and concrete Young’s

modulus and fct* is a lowered value of concrete mean tensile strength used to calculate

residual crack width. All values should be used in MPa to get crack width in millimeters.

Applying a safety factor of 10% for maximum cracks, the proposal is as follows

𝑤𝑚𝑎𝑥,𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 < 0.9𝜔𝑚𝑎𝑥

𝑤𝑚𝑎𝑥,𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 < 0.9𝜔𝑚𝑎𝑥,𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒

𝑤𝑆𝐿𝑆,𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 < 0.2𝑚𝑚

𝑤𝑟𝑒𝑠,𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 < 0.9𝜔𝑟𝑒𝑠

𝑤𝑟𝑒𝑠,𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 < 𝜔𝑟𝑒𝑠,𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒

23

Where 𝜔𝑆𝐿𝑆,𝑚𝑒𝑎𝑠𝑢𝑟𝑒𝑑 is the maximum measured crack at total applied load at the load

level that corresponds to the serviceability limit state during the load test.

24

INVESTIGATION DESIGN AND METHODOLOGY

Methodology of the investigation consists on submitting experimental results to the threshold

values from the stop criterion defined in the previous section. This way, reliability, accuracy,

level of safety and applicability to bridges of each stop criteria can be discussed Additionally,

a step by step process on how to find Vos’s proposal threshold values is explained.

Data was collected by submitting two beams cast in the laboratory to several tests and

monitoring its behavior. By monitoring the beam, data needed to apply stop criterion was

obtained with sensors.

DESCRIPTION OF EXPERIMENTS

Tests were carried at TU Delft in the Stevin II Laboratory on two beams cast in the laboratory.

The tests for this research are fully explained in the associated analysis report. (Lantsoght. E.,

2016).

Three experiments marked as P804 were carried out on a beam of 10 m long cast in the

laboratory in order to evaluate the stop criteria used during proof load testing, with material

properties designed to resemble concrete solid slab bridges. Additionally, one 8m beam

marked as P502 was tested

Beam Geometry

The cross section of the specimen P804 is 800mm x 300mm with 6 plain bars of 20 mm

each, with a total area of steel As = 1885mm2. The effective depth is dl=755mm and the

25

reinforcement ratio ρ=0.83%. The additional experiment, P502A2, was carried out on a

500 x 300mm beam with 3 plain bars of 20mm instead of 6. (Lantsoght. E, 2016)

Material properties

For concrete, the average compressive strength, obtained following the respective

standards is 63.51 MPa at 28 days with a density of 2429.6 kg/m3 tested at the age of

90 days. For the second beam, P502, the compressive strength was 71.47 MPa.

For the plain bars a yielding stress of 296.8 MPa and ultimate stress of 425.9MPa was

measured. This properties resemble the existing ones in slab bridges built in the 60’s in

the Netherlands.

Test Set up

The beam is simply supported as shown in Figure 4, subjected to a point load at a

distance a away from the support, with the values varying depending on each

experiment.

Figure 4. Beam experiments layout

26

The values from the figure for each experiment are shown in the Table Nr.1

Table 1: Values of the experiment layout.

Experiment a (mm) h (mm) I (mm) L (mm) 𝐹𝑚𝑎𝑥 (kN)

P804A1 3000 800 8000 10000 207

P804A2 2500 800 8000 10000 231

P804B 2500 800 8000 10000 196

P502A2 1000 500 5000 8000 150

Crack opening, horizontal and vertical deformation, deflection at loading and supports,

acoustic emission and strain were measured using LVDTs, laser distance finders,

acoustic emission sensors and photogrammetric measurements.

Loading procedure

Load scheme for every experiment is shown in the figure 5.

(a) (b)

27

(c) (d)

Figure 5. Experiments Loading Scheme: (a) P804A1; (b) P804A2; (c) P804B; (d) P502A2

As can be seen, in none of the eperiments the loading procedure follows the one established by

ACI 437.2M-13. As shown in Figure 2.

TEST RESULTS

Results needed for the considered stop criterion are deflection, crack width and concrete

strain. The steel strain needed for German guideline stop criterion was not measured. Given

the amount of data registered, results are show in the plots found in Figures 6 to 9. The above

mentioned parameters are plotted in terms of the force applied.

P804A1

Figure 5 shows plots resuming results obtained for this experiment. Load displacement data

can be seen on Figure 6(a). Data for displacement is the mean value of measurements from

two lasers and corrected according to the displacement support obtained from other two lasers

as well. The load displacement data is used for ACI deviation from linearity index and

permanency ratio. The deflection data is used for German guideline residual deflection criteria

28

and ACI maximum deflection and Vos’s maximum deflection proposal. Figure 6(b) shows

strain in terms of the load. This result is used to analyze concrete strain stop criteria. Figures

6(c) and 6(d) are used to analyze German guideline crack width criteria and Vos’s crack width

proposal.

(a)

29

(b)

(c) (d)

Figure 6. P804A1 experiment results: (a) load-displacement; (b) load-strain; (c) load-

crack width; (d) time-crack width

P804A2

The load displacement data as seen on Figure 7(a) was obtained the same way as experiment

P804A1. All plots from Figure 7 are analyzed in the same way as for the previous experiment.

The load versus strain plot on figure 7(b) is difficult to read given that on every cycle

measurements are very close to the previous one. To analyze the concrete strain data correctly,

the time history of the strains and loading scheme on Figure 5(b) are compared to verify the

load versus strain results. The same procedure was followed for the crack width data.

30

(a)

(b)

(c)

31

(d) (e)

Figure 7. P804A2 experiment results: (a) load-displacement; (b) load-strain; (c) time-

strain (d) load-crack width; (e) time-crack width

P804B

Experiment P804B was subjected to monotonic loading. As seen on Figure 8(a) no loading

cycles were applied, however from the loading scheme on Figure 5(c) it can be seen that the

load was held constant at certain magnitudes to take measurements. Without loading cycles no

data of residual deformation or residual crack width is available, therefore this experiment is

just analyzed using the ACI deflection criteria, and Vos’s deflection proposal taken from

displacement data in Figure 8(a), and the German guideline concrete strain stop criterion with

data from Figure 8(b). Crack width data were erratic, so that no useful information could be

taken to compare in terms of the stop criteria. Therefore the maximum crack width criterion

from German guideline and Vos’s proposal could not be analyzed. When maximum load was

reached, the sensor was outside of its measurement range, as a result of the explosive nature of

shear failure, this explains the shape of the plot at its maximum load.

32

(a)

(b)

Figure 8. P804A1 experiment results: (a) load-displacement; (b) load-

strain.

33

P502A2

Plots for experiment P502A2 as shown in figure 9 are used in the same way as for experiment

P804A1 to analyze the data and compare these to the corresponding stop criteria. In what

concerns concrete strain, the plot in figure 9(b) is difficult to read and some data overlaps.

Therefore time history of the strain n Figure 9(c) and load versus strain data from the loading

scheme on Figure 5(d) are compared to verify load versus strain data. The same procedure was

applied to verify the data from the load versus the crack width plot on Figure 9(d). On figure

9(d) it can be seen that during a part of the test, the largest crack width was measured by

LVDT 6, but later on measurements from LVDT 7 become bigger as shown on Figure 9(e).

Data from both LVDTs was used to analyze the corresponding stop criteria. However, by the

end of the test, the measurements from LVDT 6 become much larger and different from the

other measurements probably because of the opening of a crack after yielding.

(a)

34

(b)

(c)

(d) (e)

Figure 9. P502A2 experiment results: (a) load-displacement; (b) load-strain; (c) time-

strain (d) load-crack width; (e) time-crack width

For the crack width plots, since cracks appear all over the beam, four LVDT’s were placed in

different places of the specimen. Therefore, the crack width plot shows four values of the

35

crack width in terms of the force applied. The values of each LVDT vary depending on its

location.

CALULATION OF WERNER VOS STOP CRITERIA

Deflection

The diagram from figure 10 is considered for the calculations.

Figure 10. Experiment layout for Vos’ proposal calulation

Equation (8) can be rewritten as

𝛿(𝑥) = ∬𝑘 𝑑𝑥 (13)

And, based on the semi-linear approach on the moment curvature diagram describing

an elastic material, the deflection can be calculated according to

𝛿(𝑥) = ∬𝑀(𝑥)

𝐸𝐼𝑑𝑥

(14)

with its respective stiffness. Given the semi-linear assumption, the deflection only

depends on the moment M which is a function of the distance from support x. Different

36

loads can be separated with the superposition principle given the semi-linear

assumption.

A concentrated load is used in all the experiments. Given this case, the equation for the

deflection under a concentrated load is:

𝛿(𝑥) =𝑃 𝑎 (𝐼𝑠𝑝𝑎𝑛 − 𝑥)

6 (𝐼𝑠𝑝𝑎𝑛)𝐸𝐼∗(𝑥2 + 𝑎2 − 2(𝐼𝑠𝑝𝑎𝑛 𝑥))

(15)

Where x=0 at the support, P is the concentrated load at which the yielding moment

occurs and 𝐸𝐼∗ correspond to the stiffness 𝐸𝐼𝑜 or 𝐸𝐼𝑡𝑒 depending on which deflection is

being calculated. Equation (15) only applies when a < x

37

Given the load arrangement, the maximum moment occurs under the concentrated

load, but this is not the same position for maximum deflection; in Vos’s example (Vos.

W., 2016) this does not occur. Therefore P for yielding must be in accordance of the

maximum moment, which for this case is under the concentrated load, and deflection

must be calculated where it is maximum.

𝐸𝐼𝑜 is calculated with concrete modulus of elasticity and the moment of inertia of the

cross section and the reinforcement combined.

𝐸𝐼𝑡𝑒 is calculated as follows:

𝐸𝐼𝑡𝑒 = (4.91 𝜌𝑜2 + 17.66 𝜌𝑜 +

117.72

7.274 ∗ 10−4 𝑓𝑦2 + 𝜌𝑜 + 4)𝑏 𝑑3 10−7

𝑘𝑁

𝑚2

(14)

Where 𝜌𝑜 is the reinforcement ratio in percentage, 𝑓𝑦 is the steel yielding strength,

and b and d are the width and effective depth of the beam respectively. Factors have

been transformed to work with units MPa and mm.

For the maximum deflection, a residual deflection must be added. Nonetheless as

Werner Vos states, the of residual the deformation cannot be done accurately with this

method. Since it is calculated using the whole graph with the retrograde branch, the

resulting value includes errors from all the branches since a semi-linear behavior was

assumed. Therefore, it is chosen to take the residual deformation equal to zero, which

will lead to conservative results as the maximum allowable deflection is smaller. The

only deflection that must be added to δ(x) is the deflection from self-weight at the

distance x where maximum deflection occurs.

38

The yielding moment My, must be found in order to establish the value of the

concentrated load P. Vos, in his proposal uses the following formula:

𝑀𝑦 = 𝐴𝑠 𝑓𝑦 (𝑑 −1

3 (√(𝛼𝑒𝜌)2 + 2𝛼𝑒𝜌 − 𝛼𝑒𝜌) 𝑑)

(15)

Where 𝛼𝑒 is the ration between the steel and the concrete Young’s modulus and 𝜌 is

the reinforcement ratio.

In this report, the yielding moment was found using Thorenfeldt’s theory according to

the moment curvature diagram. These calculations can be found in appendix A. There

is no significant difference in the values found.

Additionally, given that the superposition principle is being considered, and 𝑀𝑦 is

caused by both concentrated and distributed loads, the load P must cause a moment

equal to 𝑀𝑦 −𝑀𝑔, where 𝑀𝑔 is the moment due to beam’s self weight, at a distance x

where maximum moment occurs.

Vos suggests to change to work in terms of the load that causes yielding, and

establishing a value of x so that the remaining equation for the deflection consists only

of a factor multiplying the moment and the stiffnes. Calculations for this report were

done in a more general way and can be found in appendix A.

Crack Width

Equations (9) and (10) are based on Van Leeuwen’s research about the influence of

crack width on corrosion of the reinforcement. According to Van Leeuwen, the crack

width can be found if the crack spacing s is known. Applying the correction from the

old notation to Eurocode notation, and taking into account that the spacing in-between

cracks is an average rather than a specific value, Vos states:

39

𝑠 = (𝑐 +1

2∅ + 0.3 𝑛∅)(1 + √

1

𝜌𝑛)

(16)

Where c is the smallest distance from a bar to a corner of the beam, ∅ is the bar

diameter and n is the number of bars in tension.

In equations (9) and (10) it can be seen that the only difference between maximum and

minimum crack width is the stress the beam is subjected to. For the maximum value,

the yielding stress is used, while the minimum crack width is calculated using the

stress in the steel caused by self-weight.

For equations (9) and (10), as stated before, β can be taken equal to 1. s is calculated

with equation (16) and σs1 is calculated according to Eurocode 2

Vos also assumes that for plain bars, the long term bond 𝜏𝑏𝑚 = 0. Therefore it does

not appear on the equation. This produces more conservative results.

For The Eurocode equations (11) and (12), the terms needed are calculated as follows:

To find the steel stress under self-weight in a cracked section the following formula is

used:

𝜎𝑠 =𝑀𝑔

𝐴𝑠 𝑧

𝑧 = 𝑑 −1

3𝑥

𝑥 = (√(𝛼𝑒𝜌)2 + 2𝛼𝑒𝜌 − 𝛼𝑒𝜌)𝑑

(17)

(17.1)

(17.2)

Where z is the lever arm.

40

𝑓𝑐𝑡𝑚 𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒 = 2.12 𝐿𝑛 (1 +𝑓′𝑐

10)MPa

𝑓𝑐𝑡𝑚 𝐴𝐶𝐼 =7.5

12√𝑓′𝑐MPa

𝜏𝑏𝑚 = 𝛼𝑓𝑐𝑡𝑚

(18)

(19)

(20)

𝑓𝑐𝑡∗ =

𝑀𝑔

𝑀𝑟 𝑓𝑐𝑡𝑚

Where α varies between 1.8 and 2. However Vos states that for plain bars this

value is around 1

(21)

𝜌𝑒𝑓𝑓 =𝐴𝑠

𝐴𝑐 𝑒𝑓𝑓

Where Ac eff is the effective area of concrete under 1tension with a height hc eff

calculated as follows:

(22)

ℎ𝑐 𝑒𝑓𝑓 = 𝑚𝑖𝑛

{

2.5(ℎ − 𝑑)(ℎ − 𝑥)

3ℎ

2

(22.1)

For the calculation of Werner Vos’ stop criteria, the following parameters according to

the cross section must be found:

o Moment of inertia of the compound section for the stiffness 𝐸𝐼𝑜 using the equivalent

concrete and steel cross section.

o Stiffness from the retrograde branch on the moment curvature diagram 𝐸𝐼𝑡𝑒 from equation

(14)

o Cracking moment𝑀𝑟, using concrete tensile from equation (18) or (19) and a linear elastic

analysis, strength from yielding moment 𝑀𝑦, with equation (15) or Thorenfeldt’s theory as

41

shown on appendix A, and moment by self weight 𝑀𝑔 using a linear elastic analysis since

no cracking occurs due to self-weight.

o Steel stress and strain under self-weight 𝜎𝑠, 𝜀𝑠 from a linear elastic analysis

o Crack spacing s with equation (16).

A comparison among theoretically determined values used for Vos’s calculation and

measured values can be seen in Table 2.

Table 2: comparison on calculated and real ultimate load

Experiment Theoretical My (kNm) Corresponding load to My(kN) Failure load (kN)

P804A1 375 177 207

P804A2 375 196 231

P804B 375 196 195

P502A2 116 139 150

42

ANALYSIS OF RESULTS

In this section,the experimental results are compared to the threshold values obtained from the

codes and Werner Vos’ proposal. The results from the four experiments are discussed based

on the previous comparison

The following measured values are the ones closest to the threshold that were taken while the

load was being held constant as can be seen on the loading scheme.

𝜀𝑐0 for this report was calculated with a linear elastic analysis since moment caused by self-

weight was less than the cracking moment. The calculations can be found on appendix A.

Comparison with Code Values:

804A1

Concrete Strain

Threshold Value: 𝜀𝑐 𝑙𝑖𝑚 − 𝜀𝑐0 = 0.0008 − 0.000031 = 0.000769 or 769 micro strain

𝜀𝑐0 was found based on the self-weigh moment 𝑀𝑔. Since cracking moment has not

occurred, a linear elastic analysis is used to find concrete existing strain:

𝜎𝑐0 =𝑀𝑔 𝑦

𝐼

Table 3: concrete strain measured on P804A1 at every load step

After 90kN which represents 43% of the ultimate load of 207 kN, the limit for the concrete

strain was reached.

Load (kN) εc (με)

75 570

83 707

90 820

118 1200

136 1400

158 1600

179 1800

43

DASfB Deflection:

Threshold: Increase of 10% in residual deflection.

The stop Criterion was exceeded in load level number 2 for a load of 90kN which is 34%

of the failure load. After the load increase from 90kN to 120kN a residual deflection

increase of 36% was measured, exceeding the threshold value of 10%.

DAfStB crack width:

Threshold: 0.5mm for the width of the maximum crack, 30% of maximum crack for

residual crack width.

Table 4: P804A1 crack width. From maximum and minimum among all LVDTs

Load (KN) 𝑤𝑚𝑎𝑥 (mm) 𝑤𝑟𝑒𝑠 (mm) 0.3 𝑤𝑚𝑎𝑥 (mm) 75 0.00 0.00 0.00

84 0.00 0.00 0.00

119 0.17 0.04 0.05

139 0.28 0.08 0.08

158 0.42 0.14 0.13

178 0.55 0.22 0.16

Since this was the first experiment carried out on the beam, threshold values are

considered for newly caused cracks. As can be seen on Figure 6(c), only LVDT 13 and 15

performed correctly. Values from other sensors are erratic. Cracks smaller than 0.05mm

should not be considered (Lantsoght, E., 2016). The resuls in Table 4 come from LVDT 15

which showed the largest values. It can be seen that the threshold value for the maximum

crack width was reached on the increment to the 175kN step, close to the ultimate load. As

for maximum residual crack width, it was reached one load step earlier on the cycles of

160 kN.

44

ACI Deviation from linearity index

Threshold: 0.25

Table 5: P804A1 Linearity index

Points from load-deflection plot to obtain α were taken at the maximum load on the first

cycle of each increment where the load was held constant. During the load increment from

75kN to 85kN after ten cycles of 75kN load, a value for IDL of 0.37 which surpasses the

threshold value of 0.25 was found. A load of 75kN represents 36% of the ultimate load of

207KN. For every other value of α taken from the plot, the threshold was also exceeded.

ACI Permnency ratio

Threshold: Ipr < 0.50

Table 6: P804A1 permanency ratio

Load (kN) ∆𝑟 (mm) ∆𝑚𝑎𝑥 (mm) I I pr

75

0.16 1.76 0.09 0.47

0.08 1.81 0.05 1.95

0.18 1.95 0.09

120

0.09 4.49 0.02 0.063

0.01 4.60 0.00 14.96

0.09 4.67 0.02

tan αref 19.91 𝐼𝐷𝐿 tan α1 12.57 0.37

tan α2 11.68 0.41

tan α3 10.69 0.46

tan α4 10.05 0.49

tan α5 9.88 0.50

45

Given the variation of the results and the fact that the threshold was exeeded in the first

cycle of 75kN, no more values were taken from the plot for the calculation of permanency

ratio.

ACI maximum deflection

Based on the span length the threshold value is:

∆𝑙 ≤𝐿

180=8000

180= 44.4 𝑚𝑚

Table 7: P804A1, deflection measured at every step

Load (kN) ∆𝑙 (mm) 75 3.06

85 5.12

120 8.31

140 10.46

158 12.51

179 14.18

On the last load increment before yielding occurred, the deflection of the beam is far

from reaching the threshold value. In Figure 6(a) it can be seen that even after yielding

and removing the load, the maximum deflection was around 24mm, which is much less

than 44mm. For the residual deflection, ACI 437.2M-13 states that it must be less that

¼ of the maximum deflection. However this value must be measured 24 hours after the

test, which was not the case for this experiment.

P804A2

Concrete Strain

Threshold: 0.000771

Table 8: concrete strain measured on P804A2 at every load step

46

Load (kN) εc (με)

74 434

116 0.741

158 1050

197 1300

215 1490

231 1600

The concrete strain limit value was reached at 120kN, which is 51.7% of the ultimate

load.

DAfStB deflection

At the second level of loading and increae of 18% of rpermanent deformation was

measured. The stop criterion was exceeded for a load of 120 kN, which is 51% of the

maximum load.

DAfStB crack width

Threshold: 0.3mm for the width of the maximum crack, 20% of maximum crack for

residual crack width.

Table 9: P804A2 crack width. From maximum and minimum among all LVDTs

Load (KN) 𝑤𝑚𝑎𝑥 (mm) 𝑤𝑟𝑒𝑠 (mm) 0.2 𝑤𝑚𝑎𝑥 (mm) 75 0.12 0.00 0.024

115 0.21 0.0182 0.042

159 0.31 0.0186 0.062

192 0.41 0.0222 0.082

232 0.50 0.0036 0.1

Figure 7(d), shows the crack width of every LVDT against the applied load. However

it is difficula to read the data. Nonetheless when comparing the plots of crack width

versus time and load versus time in figures7(d) and 7(e) respectively, it can be seen

that LVDT 15 measured the largest values. These values are used for Table 9. Since

47

P804A2 was the second test on the beam, threshold values were taken for existing

cracks. When increasing the load to 160kN, the maximum crack width of 0.3mm was

reached. The stop criterion for the residual crack width was not exceeded until the

ultimate load of 231kN was applied, even when residual values where measured under

a load of 10kN.

ACI Deviation from Linearity Index

Threshold: 0.25

Table 10: P804A2 Linearity index

tan αref 23.41 1-tan α /tan αref

tan α1 21.84 0.07

tan α2 21.26 0.09

tan α3 20.16 0.14

tan α4 19.08 0.19

The value of tan(α) in Table 10 represents the slope at a point of the load-displacement

plot. As seen in Figure 8(a) these slopes are very similar to each other in every cycle,

which suggests that deviation from linearity index is small. Indeed, as seen in table 10,

the threshold value of 0.25 is not reached even in the last load step when the beam

failed

ACI Permanency Ratio

Threshold: Ipr < 0.5

48

Table 11: P804A2 permanency ratio

Load (kN) ∆𝑟 (mm) ∆𝑚𝑎𝑥 (mm) I Ipr

75

0.038 2.882 0.013 -0.227

-0.008 2.828 -0.003

115

0.046 4.855 0.009 0.414

0.019 4.907 0.004

160

0.159 7.141 0.022 -0.050

-0.008 7.063 -0.001

197

0.058 9.042 0.006 3.647

0.212 9.081 0.023

The values of the residual and maxium deformation needed for calculations were taken by

comparing load versus time data and deflection versus time, since the load-deflection plot

was hard to read and only considers values on the same load level. As it can be seen in

table 11, the values of permanecy ratio vary from much smaller to much larger the

threshold value of 0.5, and even negative values were found. This negative values occurs

when residual deformation was smaller than in the previous cycle. Therefore it can’t be

determined when this stop criterion was exceeded.

ACI maximum deflection

Threshold value 44.4 𝑚𝑚

Table 12: Deflection P804A2 at every load step

Load (kN) ∆𝑙 (mm) 75 3.20

115 5.32

159 7.56

96 9.58

225 12.97

49

As in experiment P804A1, this acceptance criterion was never exceeded. The

maximum deflection value after shear failure was around 13mm.

P804B Monotonic Load

Stop criteria not considered for this experiment are omitted

Concrete Strain

Threshold: 0.000771 or 771 microstrain

Table 13: concrete strain measured on P804B at every load step

Load (kN) εc (με)

72 106

82 138

92 351

110 821

120 912

15 1200

186 1400

195 1453

At a load of 105KN or 53% of the ultimate load, the stop criterion for the concrete

strain is exceeded.

Crack Width

Since this experiment was carried following a monotonic loading protocol, there

are no values of the residual crack width to compare with the stop criterion. As for

the maximum crack width, the collected data were erratic and could not be

analized.

ACI Maximum and Minimum Deflection

50

Threshold: max=44.1mm; min:25% of maximum measured

Table 14: deflection measured on P804B at every load step

Load (kN) ∆𝑙 (mm) 70 2.57

83 3.51

112 5.93

120 7.27

149 8.65

158 9.23

171 10.58

192 11.84

9 22.89

Even after shear failure, the maximum deflection from the acceptance criterion was not

reached. The residual minimum deflection at was masured at the very end of the test,

not 24 hours later as stated in the code. Moreover the residual deformation large value

is an effect of shear crack developed at failure. Therefore ACI 437.2M-13 residual

deflection stop criterion can’t be compared in this experiment.

P502A2

Concrete Strain

Threshold= Threshold Value: 𝜀𝑐 𝑙𝑖𝑚 − 𝜀𝑐0 = 0.0008 − 0.0000089 = 0.000791

Table 15: concrete strain measured on P502A2 at every load step

Load (kN) εc (με)

48 342

73 471

97 831

121 1000

121 1100

146 1200

138 3300

51

At a load of 91kN the stop criterion was exceeded. This load representes 61% of the

ultimate load of 150kN

DAfStB defletion

In this experiment the scpecimen is unloaded to 0kN on every cycle, therefore

measured residual deflections are very small and small changes would imply an

increase grater then 10% from the original premanent deflection> however at all the

unloading steps down to 0kN cthe stop criterion was never exceeded.

DAfStB crack width

Threshold: 0.3mm for maximum crack, 20% of maximum crack width for residual

cracks.

Since maximum values for crack width vary between two LVDTs, both values are

presented.

Table 16: P502A2 crack width. Maximum and residual cracks as measured by LVDTs

Load

(KN)

𝑤𝑚𝑎𝑥 (mm) LVDT 6

𝑤𝑚𝑎𝑥 (mm)LVDT 7

𝑤𝑟𝑒𝑠 (mm)LVDT 6

𝑤𝑟𝑒𝑠 (mm)LVDT 7

0.2 𝑤𝑚𝑎𝑥 (mm)LVDT6

0.2 𝑤𝑚𝑎𝑥 (mm)LVDT7

47 0.079 0.034

72 0.142 0.061 0.0003 0.0002 0.028 0.012

97 0.212 0.089

122 0.265 0.168 -0.001 0.027 0.053 0.033

122 0.261 0.184

146 0.340 0.364 0.020 0.126 0.068 0.073

139 2.063 0.416 - 0.161 0.412 0.083

As in experminet P804A2, the crack width versus load plot in Figure 9(d) is hard to

read, but from Figure 9(e) that shows the crack width versus time, it can be seen that at

different stages of the loading, different LVDTs measured the maximum values. Table

52

16 shows the measurements from LVDTs 6 and 7. However from Figure 8(e) it can be

seen that on the last step, LVDT 6 shows larger values due to the opening of a yielding

crack as explained before. Cosidering these factors, the threshold of 0.3mm for the

maximum crack width of existing cracks was reached in LVDTs 6 and 7 between

120kN and 150 kN, at more than 80% of the ultimate load durig the last step before

reaching this maximum load. On the same step, the maximum residual crack width

value was exceeded but only by the cracks measured by LVDT 7.

ACI Deviation from linearity index

Table 17: P502A2 Linearity index

tan αref 33.19 1-tan α /tan αref

tan α1 30.09 0.09

tan α2 27.40 0.17

tan α3 15.86 0.52

The value of tan a3 was taken on the last load increase before yielding occurred at

150kN. It was not until this phase that the threshold value was surpassed.

ACI permanency Ratio

Threshold: Ipr < 0.5

Table 18:P502A2 permanency ratio

Cycle Δr Δmax I Ipr

1; 75kN 0.05 2.18 0.03 0.29

2; 120kN 0.03 3.94 0.01 5.80

3; 147kN 0.22 5.19 0.04

Based on the results from Table 18, it can be concluded that the permanency ratio

fluctuates and that this acceptance criterion cannot be used for concrete bridges.

ACI Maximum deflection

53

Threshold value 27.7 𝑚𝑚

Table 19: deflection measured on P502A2 at every load step

Load (kN) ∆𝑙 (mm) 73 2.21

97 3.04

122 4.05

147 5.29

139 8.77

The threshold value for this experiment is lower than the others because the span

length was smaller. Still, stop criterion was never exceeded. Maximum deformation

after yielding and maintaining the load was close to 9mm.

Comparison with Vos’s Proposal

Deflection

The values of the deflection were taken at the distance x from the support where deflection is

maximum.

Table 20: Comparison of deflection values from Vos’s proposal

Werner Vos Deflection Closest measured % of ultimate load

Δmin (mm) Δmax(mm) Δ(mm) Load (kN)

804A1 3.83 10.86 10.46 140 67%

804A2 3.81 10.79 9.58 195 84%

804B 3.81 10.79 10.58 171 87%

502A2 1.69 6.16 5.29 150 100%

Experiment P804A1 reached maximum deflection at a lower percentage of the ultimate load,

while P502A2 reached right after yielding occurred. P804A2 and P804B presented good

results where the maximum allowable deflection was not too conservative and a considerable

amount of the ultimate load was applied. The difference between these two experiments and

54

P804A1, is that for A1 the beam was new, with no previous loads or cracks, had never been

subjected to yielding and the failure mode was different. Since the slope of the retrograde

branch EIte represents stiffness after the yielding moment My has occurred which was not the

case for P804A1 since it was a new beam, this may explain why the stop criterion was

exceeded long before failure was reached.

As for P502A2, concentrated load was 1 m away from the support at a 5 m span, this would

cause less moments and deflection than a load located at mid-span. This reduction on

deflection due to the position of the load may have led to reaching maximum allowable

deflection after yielding. Even though the load was close to the support, on this test flexural

failure was achieved, perhaps due to the effect of previous cracking

Crack Width

In table 21, the values of maximum and residual crack width closest to the ones obtained with

Vos’s proposal are shown

Table 21: Comparison of crack width values from Vos’s proposal

closest measured WERNER VOS LIMITS

Load

(kN) 𝑤𝑚𝑎𝑥 (mm)

𝑤𝑟𝑒𝑠 (mm)

0.9𝑤𝑚𝑎𝑥,𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒 (mm)

𝑤𝑟𝑒𝑠,𝐸𝑢𝑟𝑜𝑐𝑜𝑑𝑒 (mm)

0.9𝑤𝑚𝑎𝑥,𝐿𝑒𝑒𝑢𝑤𝑒𝑛 (mm)

𝑤𝑟𝑒𝑠,𝐿𝑒𝑒𝑢𝑤𝑒𝑛 (mm)

P804A1 118 0.17 0.043 0.1829 0.00707 0.1467 0.0694

P804A2 115 0.21 0.018 0.1829 0.013 0.1467 0.0694

P804B - - - 0.1829 0.013 0.1467 0.0694

P502A2 122 0.265 0.02 0.234 0.00669 0.171 0.0075

Results are not as consistent as the ones obtained with the German guideline crack width. For

P804A1, the maximum crack width was reached at a 57% of the maximum load for the code

55

threshold and it already surpassed Van Leuween’s threshold. In P804A2 the stop criterion was

exceeded at 49% of the maximum load, while for P502A2 the stop criterion was exceeded at

81% of the ultimate load both for the maximum crack width of the Eurocode and Van

Leeuwen’s crack width. The stop for the residual crack width on the other hand was surpassed

only for the value obtained from the Eurocode on experiments P804A1, 804A2 and P502A1.

Therefore results are not consistent and a clear relation between conditions of the specimen

and the results can be done. On what concerns failure mode, a relation cannot be established

either. Experiments P804A1 and P52A2 presented flexural failure but the fraction of the

ultimate applied load at which the threshold value was exceeded is considerably different

probably because of the presence of cracks on experiment P502A2. As for P804A2 which

failed under shear, percentage of the ultimate load applied at the moment the criterion was

surpassed is even less than in the other two experiments

DISCUSSION

In this section, comparisons made in the previous section will be analized for every stop

criterion:

Concrete strain

Table 22 summarizes the results on every experiment for this criterion

Table 22: Comparison of concrete strain on every experiment

MEASURED DAfStB

Load KN μεc εc (με) 𝜀𝑐 𝑙𝑖𝑚 (με) % of ultimate load

804A1 90 790 769 43%

804A2 120 770 767 51%

804B 105 790 767 53%

502A2 91 810 791 61%

56

As can be seen, on all experiments, stop criteria was reached between 40% and 60% of

the ultimate load. P804A1, in which the stop criterion was exceeded at the lowest

fraction of the ultimate load was the only test where the beam was un-cracked besides

P804B where the un-cracked part was tested. As for P804A2 and P804B, both reached

the stop criterion at about half the ultimate load. Between these two experiments

differences where the loading protocol, and that for P804B part of the beam was un-

cracked which was the part that was tested. Differences between these two experiments

and P804A1 was a variation of 500mm of the position of the load and the conditions of

the beam prior the test. In an 8m span, one would expect that a 0.5m change in position

of the load may not have a big influence. However, this small variation in the position

of the load was large enough to change the failure mode from flexure to shear, which

consequently influenced the results at which stop criteria was reached . P502A2 got

the closest to the ultimate load before exceeding the stop criterion. This was a cracked

beam, but the span length was smaller and the position of the load was closer to the

support.

Additionally, when finding 𝜀𝑐 𝑙𝑖𝑚, 𝜀𝑐0 which is existing strain from permanent loads

must be considered, but the limit value does not consider conditions of the beam or

load history that may change the value of 𝜀𝑐0. For example, in bridges, continuous

traffic which is not a permanent and will not be present during the test may have an

influence on existing strain, but it is not considered in calculations, thus, 𝜀𝑐0 is

miscalculated, setting a wrong threshold value for this stop criteria. Overall, this stop

criterion shows consistent results. However, even a though 60% of the ultimate load is

57

considered a good fraction of the load to be applied, 40% which was the case for

experiment P804A1 may be too conservative to stop the test since stop criteria was

exceeded. Nonetheless, P804A1 does not represtent the actiual conditions of beams on

a bridge. Therefore, overall concrete strain seems a reliable stop criteria to apply on

bridges

DAfStB Deflection

One would expect to reach limit for the residual deflection as the structure gets closer

to failure, as for the case of 804A2 However for P804A1 and P502A2 the increment

occurred more than once and it occurred in the first cycles at which the structure is far

from suffering permanent damage, and stopping the test at that point would not provide

any relevant information.

Additionally, on the first cycles, the residual deformation is really small, especially if

load is decreased until 0kN as in experiment P502A2. Therefore, small variations

would represent a 10% increment, which may be the reason of the +10% increment in

the first steps. Perhaps, loading protocol should not decrease the load to 0kN, or

residual deformation should be measured at an established load before reaching 0kN.

Also, generally, when proof load tests are performed, a baseline load level is

maintained to keep sensors ad jacks activated during the test

58

Comparing experiments, results are not consistent since in some cases stop criteria was

exceeded and in some it was not.

Crack Width

Table 23 shows the summarized results of crack width on every experiment

Table 23: Comparison of maximum crack width on every experiment

DASfB limits Closest measured % of ultimate load

𝑤𝑚𝑎𝑥 (mm) 𝑤𝑟𝑒𝑠 (mm) Load (kN) 𝑤𝑚𝑎𝑥 (mm) 804A1 0.50 0.17 178 0.56 86%

804A2 0.30 0.09 159 0.31 69%

804B 0.30 - - -

502A2 0.30 0.05 123 0.26 81%

Table 24: Comparison of minimum crack width on every experiment

DASfB limits Closest measured % of

ultimate

load

𝑤𝑚𝑎𝑥 (mm) 𝑤𝑟𝑒𝑠 (mm) Load (kN) 𝑤𝑚𝑎𝑥 (mm) 𝑤𝑟𝑒𝑠 (mm) 804A1 0.50 0.083 139 0.28 0.082 67%

804A2 0.30 0.06 159 0.31 0.02 69%

804B 0.50 - - - -

502A2 0.30 0.07 146 0.364 0.12 97%

In what concerns the maximum crack width, for experiments P804A1 and P502A2, the

maximum crack width was reached at around 85% of the ultimate load, whereas for

P804A2 it was around 70%. For the three cases, a considerable amount of the

maximum load had already been applied, and considering a safety margin to avoid

permanent damage, threshold values seem adequate for the stop criteria. Also, the

threshold value, considers if cracks are new or already existing. Therefore, maximum

crack width provides consistent results and it is not as conservative as the results

obtained for concrete strain, allowing the test apply greater loads on the structure so

that relevant information can be obtained

59

As for residual crack width, experiment P804A1 surpassed the threshold value but it

happened on a cycle before reaching the maximum crack width as shown on Tables 24

and 24 the maximum crack width values was surpassed at a load close to 180kN while

the value of maximum residual crack width was reached after applying a load of

140kN. For P502A2 on the other hand, the residual crack width stop criterion was

exceeded one cycle before failure really when the maximum crack width threshold had

already been exceeded. As for P804A2 this threshold was not surpassed before failure

occurred. Table 24 shows the closest measured value to the threshold from the stop

criteria, however, this is still far from being reached.

When analyzing residual crack width it should be considered that aggregate that spalls

as concrete breaks, could get stuck inside cracks and avoid its closure while load is

being removed causing a residual crack width larger than the allowable. For this case,

the larger reading does not signal irreversible damage to the structure. Residual crack

width also presented problems when collecting the data. As can be seen on the Figure

7(d), some measured values of residual crack width are negative, mostly because these

are very small values and measurements combine elastic deformation and crack width

which can’t be isolated. However, it should be noted that residual cracks smaller then

0.05mm can be neglected since it is considered a microcrack which is not structural

(Lantsoght, E., 2016).

Even though the stop criterion related to the residual crack width was full-filled in two

of the three experiments, the issues when measuring, and possible causes than prevent

cracks from closing should be considered when using this stop criteria.

60

ACI Deviation from linearity Index and permanency ratio

In what concerns the deviation from linearity index, P804A1 did not remained under

the threshold value in any of the load steps. Meanwhile P502A2 surpassed the

threshold at the third load step, and P805A2 remained always under the maximum

value. This can easily been seen on Figure 7(a) for experiment 804A2 which shows the

load-deflection plot, since the slopes are similar on every load step contrary to P804A1

and P502A2 force displacement plots, as shown in figure 6(a) and 9(a).

As for permanency ratio, for all three experiments the results are erratic. For one set of

cycles Ipr is much less than the threshold value and for the next set of cycles it is much

larger. Therefore this stop criterion should not be recommended for the use with proof

load testing.

These two criteria are based on taking points from the load-deflection diagram

obtained from the experiments and thus are influenced by the loading protocol that is

followed. ACI 437.2M-13 defines a loading protocol in order to apply this acceptance

criteria, shown in Figure 2. This protocol was not followed on any of the experiments

as can be seen on the loading schemes on Figure 5. Therefore, this criterion does not

provide any relevant information to this report.

The values for 𝛼𝑖 and 𝐼𝑝𝑟are sensitive to small changes on the values taken from the

load displacement plot for calculations. Since this plot depends on the loading

protocol, following ACI 437.2M-13 would provide better data to apply this stop

criteria, Otherwise, calculated values are erratic and are not useful to compare with the

threshold value. Additionally, to follow this loading protocol and it would be necessary

61

to use force controlled loading instead of displacement controlled loading, but

displacement controlled loading is safer when testing for bridges because once yielding

is reched no more force is applied.

Finally, the deviation from linearity index and permanency ratio are not exactly stop

criteria but acceptance criteria. This means that if the structures remains under the

threshold values it passes the test, but this does not explicitly means that there has not

occurred permanent damage. In experiment P804A2 for example, at all the cycles

linearity index always remained below the limit which means it passed; but it still

remained under the limit even in the last increment prior to failure which means the

beam passed the acceptance criteria but was really close to reaching permanent

damage. On the other hand, in experiment P804A1 threshold was already surpassed in

the first load cycle, which means it did not pass the proof load test under ACI

acceptance criteria. However, it was far from reaching failure.

ACI deflection

ACI 437.2M-13 defines a monotonic loading protocol in accordance to deflection

acceptance criteria which was not followed for any of the experiments, which may lead

to incongruent results. One of the main differences in the loading protocol is that the

one in ACI 437.2M-13 takes much more time: around two days to complete the test. In

this case the whole loading process was completed in less than two hours.

As can be seen in the results, the maximum allowable deflections for beams are much

larger than the maximum deflection measured prior to yielding. Even after yielding,

62

the maximum allowable values were not reached. Under this circumstances 804B beam

passed the proof load test under the maximum deflection acceptance criteria. However,

it not only was subjected to permanent damage, it actually failed, which is what is

trying to be avoided. Same happened with experiments under the cyclic loading

protocol. Even after failure occurred, the threshold value was not reached. After failing

the beam did not fullfill the criterion according to the residual deflection. However,

residual deflection value was taken at the very end of the test after permanent damage

occurred from the shear failure, so this stop criterion can’t be compared with the

measured value. Additionally, even if the beam had not failed ACI437.2M-13

establishes that residual deformation must be measured 24 hour after the load has been

removed, not right after the load is removed. On a real life scenario a building can

easily be closed for 24 hours until all the measurements are completed, but this is not

practical for bridges. Setting up the equipment, performing the test and waiting 24 hour

after the test is completed to measure residual deflection implies closing part of a road

or highway long enough to cause problems with traffic.

Werner Vos’s Deflection

In the three P804 experiments, the maximum deflection was reached before failure.

However in P804A1, the stop criterion was exceeded within a wide range before

reaching the yielding point at about 67% of the ultimate load, whereas on the other two

cases it was reached at a 87% of the ultimate load which may be considered as too

close to the yielding point or not. This difference among experiments in the fraction of

63

the ultimate load at which the threshold was reached may have to do with the failure

mode On the other hand, for P502A2 the maximum allowable deflection occurred after

yielding. Based on this, the semi-linear assumption on the moment curvature diagram

and its respective stiffness from the retrograde branch seem to be quite a good

approximation that considers the effect of cracking on the change of stiffness, it may

not be too conservative but still it occurred before permanent damage in three of the

four experiments. Additionally, the maximum values found are much smaller than the

maximum deflection found based on ACI acceptance criteria given that stop criteria

and acceptance criteria are have a slightly different function as explained before.

According to the results, there is no explicit relation in the load at which threshold was

reached and the conditions of the beam before the test. Four cases can be considered

based on conditions of the specimen and failure mode: flexural failure un-cracked like

P804A1, shear failure cracked like P802A2, shear failure un-cracked like P804B and

flexural failure cracked like P502A2. P804A1 reached the limit at a 67% of the

maximum load, which is a good fraction of the ultimate load, but does no resemble

conditions of a bridge. P804A2 was cracked, P804B was un-cracked and both failed in

shear but did it around 85% of the maximum load. However, P502A2 which was

cracked as well, reached the limit right after yielding. Based on this, a preloaded

cracked condition of the beam cannot be directly related on how it affects deflection.

Additionally, it is interesting to compare P804A2 and P804B since these two

experiments had the same load arrangement and span length therefore, calculations for

the threshold values gave the same numbers. The only difference besides loading

64

protocol was that about 1.5 m of the beam on P084B had not been tested since it was

flipped for this tests. However, results are pretty much the same between both which

means this partially un-cracked part of the beam, and the loading protocol had nothing

to do with the magnitude at which the threshold value was reached. A cyclic load may

cause fatigue and affect the stiffness in a different way a monotonic load does; and the

loading scheme has no influence when calculating EIte However, given the results

obtained between P804A2 and P804B, EIte was accurate enough to provide a maximum

value for this proposal of stop criteria.

Werner Vos’ Crack width

As can be seen, the values for the maximum crack width obtained from the Eurocode

and the Van Leuween formulas are smaller than those established in the German code,

which means Vos’s crack width limits in all experiments were reached long before

failure. For P804A1 and P804A2 it was around 57% and 49% of the ultimate load

respectively. Nonetheless for P504A2 it was around 80% of the ultimate load, which

may bbe n quite large and not conservative enough . This experiment under the

German guideline stop criteria reached the maximum value right at yielding therefore

Vos’s proposal on maximum crack width presented better results just for experiment

P502A2, but overall, German guideline maximum crack width stop criteria was more

consistent and showed more conservative results.

It should be noted that residual crack from the Eurocode for all experiments are smaller

than 0.05mm which was the magnitude at which crack width can be neglected. These

equations for calculating these values were not taken directly from their respective

65

documents, but from Vos’s work. In his proposal, the description on how to apply this

stop criteria proposal, or how to calculate threshold values is not clear. For some

formulas, the terms included are not defined clearly and in some cases input units do

not match units of the results, therefore transformation factors were included for

calculations in this report. Finally, Equation (16) for crack spacing s was not used

given that just by using the formula without any further explanation values around

400mm and 500m were found which are too large for crack spacing. s was calculated

using a graph found on Figure 12, established by Van Leuween (Van Leuween. J.,

1962) in which reinforcement ratio is related to crack spacing.

Figure 12: Average crack spacing according to Van Leeuwen (1962)

66

CONCLUSIONS

By analyzing all the theoretical and experimental data, and compare it to the stop criterion, it

can be concluded that:

The ACI 437.2M-13 cyclic loading acceptance criteria must be used only if the ACI loading

protocol is applied, otherwise the deviation form linearity index and permanency ratio values

obtained will be erratic and can’t be compared to threshold values established by the

acceptance criterion.

The ACI 437.2M-13 maximum deflection is too permissive. All specimens failed before

reaching the threshold value, and even after failure, the limit was not exceeded. The ACI

437.2M-13 residual deflection is not considered since measurements were not done 24 hours

after the test as the code requires and this procedure is not suitable for bridges.

The German guideline concrete strain stop criterion is suitable for cracked and non-cracked

beams. Its results are consistent. On cracked beams, which resemble the conditions of beam on

bridges, the stop criterion almost at 60% of the ultimate load which is a considerable amount

of the load but not too permissive. Therefore German guideline concrete strain seems to be a

good criterion to be applied on bridges.

The German guideline residual deflection stop criterion does not seem to be a suitable

criterion for both cracked and non-cracked beams. Results are not consistent and in two

experiments the threshold value was surpassed during the first cycles. This would lead to

67

cancelation of a proof load test before obtaining any relevant information from the structure,

and it would require closing of a structure that is still suitable for use.

The German guideline crack width stop criteria provided good results in what concerns the

stop criterion for the maximum crack width, where about 80% of the maximum load was

applied. This criterion was less conservative than the concrete strain stop criterion and does

distinguish its threshold values between cracked and non-cracked specimens. The stop

criterion seems to be a good criterion to apply regardless of the loading protocol that is being

followed. The residual crack width on the other hand, did not provide results as consistent as

the maximum crack width. When applying this stop criterion, it should be considered the

aggregate preventing cracks from closing, and cracks smaller than 0.05mm should be

neglected. Finally, positioning of the sensors should be done carefully in order to take

measurements correctly, especially when dealing with non-cracked beams, where location of

formation of cracks